Face Generation¶

In this project, you'll use generative adversarial networks to generate new images of faces.

Get the Data¶

You'll be using two datasets in this project:

MNIST
CelebA

Since the celebA dataset is complex and you're doing GANs in a project for the first time, we want you to test your neural network on MNIST before CelebA. Running the GANs on MNIST will allow you to see how well your model trains sooner.

If you're using FloydHub, set data_dir to "/input" and use the FloydHub data ID "R5KrjnANiKVhLWAkpXhNBe".

data_dir = './data'

"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
import helper

helper.download_extract('mnist', data_dir)
helper.download_extract('celeba', data_dir)

Found mnist Data
Found celeba Data

Explore the Data¶

MNIST¶

As you're aware, the MNIST dataset contains images of handwritten digits. You can view the first number of examples by changing show_n_images.

show_n_images = 25

"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
%matplotlib inline
import os
from glob import glob
from matplotlib import pyplot

mnist_images = helper.get_batch(glob(os.path.join(data_dir, 'mnist/*.jpg'))[:show_n_images], 28, 28, 'L')
pyplot.imshow(helper.images_square_grid(mnist_images, 'L'), cmap='gray')

<matplotlib.image.AxesImage at 0x7f4bf5224978>

CelebA¶

The CelebFaces Attributes Dataset (CelebA) dataset contains over 200,000 celebrity images with annotations. Since you're going to be generating faces, you won't need the annotations. You can view the first number of examples by changing show_n_images.

show_n_images = 25

"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
mnist_images = helper.get_batch(glob(os.path.join(data_dir, 'img_align_celeba/*.jpg'))[:show_n_images], 28, 28, 'RGB')
pyplot.imshow(helper.images_square_grid(mnist_images, 'RGB'))

<matplotlib.image.AxesImage at 0x7f4bfc6a50b8>

Preprocess the Data¶

Since the project's main focus is on building the GANs, we'll preprocess the data for you. The values of the MNIST and CelebA dataset will be in the range of -0.5 to 0.5 of 28x28 dimensional images. The CelebA images will be cropped to remove parts of the image that don't include a face, then resized down to 28x28.

The MNIST images are black and white images with a single color channel while the CelebA images have 3 color channels (RGB color channel).

Build the Neural Network¶

You'll build the components necessary to build a GANs by implementing the following functions below:

model_inputs
discriminator
generator
model_loss
model_opt
train

Check the Version of TensorFlow and Access to GPU¶

This will check to make sure you have the correct version of TensorFlow and access to a GPU

"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
from distutils.version import LooseVersion
import warnings
import tensorflow as tf

# Check TensorFlow Version
assert LooseVersion(tf.__version__) >= LooseVersion('1.0'), 'Please use TensorFlow version 1.0 or newer.  You are using {}'.format(tf.__version__)
print('TensorFlow Version: {}'.format(tf.__version__))

# Check for a GPU
if not tf.test.gpu_device_name():
    warnings.warn('No GPU found. Please use a GPU to train your neural network.')
else:
    print('Default GPU Device: {}'.format(tf.test.gpu_device_name()))

TensorFlow Version: 1.8.0
Default GPU Device: /device:GPU:0

Input¶

Implement the model_inputs function to create TF Placeholders for the Neural Network. It should create the following placeholders:

Real input images placeholder with rank 4 using image_width, image_height, and image_channels.
Z input placeholder with rank 2 using z_dim.
Learning rate placeholder with rank 0.

Return the placeholders in the following the tuple (tensor of real input images, tensor of z data)

import problem_unittests as tests

def model_inputs(image_width, image_height, image_channels, z_dim):
    """
    Create the model inputs
    :param image_width: The input image width
    :param image_height: The input image height
    :param image_channels: The number of image channels
    :param z_dim: The dimension of Z
    :return: Tuple of (tensor of real input images, tensor of z data, learning rate)
    """
    inputs_real = tf.placeholder(tf.float32, shape=(None, image_width, image_height, image_channels))
    inputs_z = tf.placeholder(tf.float32, shape=(None, z_dim))
    learning_rate = tf.placeholder(tf.float32, shape=())
    
    return inputs_real, inputs_z, learning_rate

"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_model_inputs(model_inputs)

Tests Passed

Discriminator¶

Implement discriminator to create a discriminator neural network that discriminates on images. This function should be able to reuse the variables in the neural network. Use tf.variable_scope with a scope name of "discriminator" to allow the variables to be reused. The function should return a tuple of (tensor output of the discriminator, tensor logits of the discriminator).

def discriminator(images, reuse=False):
    """
    Create the discriminator network
    :param images: Tensor of input image(s)
    :param reuse: Boolean if the weights should be reused
    :return: Tuple of (tensor output of the discriminator, tensor logits of the discriminator)
    """
    alpha = 0.2
    
    with tf.variable_scope('discriminator', reuse=reuse):
        # Convolution Layer 1: 28x28x3
        x1 = tf.layers.conv2d(images, 64, 5, strides=2, padding='same', kernel_initializer=tf.contrib.layers.xavier_initializer_conv2d())
        relu1 = tf.maximum(alpha * x1, x1)
        
        # Convolution Layer 2
        x2 = tf.layers.conv2d(relu1, 128, 5, strides=2, padding='same', kernel_initializer=tf.contrib.layers.xavier_initializer_conv2d())
        bn2 = tf.layers.batch_normalization(x2, training=True)
        relu2 = tf.maximum(alpha * bn2, bn2)
        drop2 = tf.layers.dropout(relu2, rate=0.2)

        
        # Convolution Layer 3
        x3 = tf.layers.conv2d(drop2, 256, 5, strides=2, padding='same', kernel_initializer=tf.contrib.layers.xavier_initializer_conv2d())
        bn3 = tf.layers.batch_normalization(x3, training=True)
        relu3 = tf.maximum(alpha * bn3, bn3)
        drop3 = tf.layers.dropout(relu3, rate=0.2)
        
        # Flatten
        x4 = tf.reshape(drop3, (-1, 4 * 4 * 256))
        logits = tf.layers.dense(x4, 1)
        out = tf.sigmoid(logits)
        
    return out, logits

"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_discriminator(discriminator, tf)

Tests Passed

Generator¶

Implement generator to generate an image using z. This function should be able to reuse the variables in the neural network. Use tf.variable_scope with a scope name of "generator" to allow the variables to be reused. The function should return the generated 28 x 28 x out_channel_dim images.

def generator(z, out_channel_dim, is_train=True):
    """
    Create the generator network
    :param z: Input z
    :param out_channel_dim: The number of channels in the output image
    :param is_train: Boolean if generator is being used for training
    :return: The tensor output of the generator
    """
    alpha = 0.2
    
    with tf.variable_scope('generator', reuse=not(is_train)):
        # Fully Connected Layer 1: 7x7x1024
        x1 = tf.layers.dense(z, 7 * 7 * 1024)
        x1 = tf.reshape(x1, (-1, 7, 7, 1024))
        x1 = tf.layers.batch_normalization(x1, training=is_train)
        x1 = tf.maximum(alpha * x1, x1)
        
        # Convolution Layer 2
        x2 = tf.layers.conv2d_transpose(x1, 512, 5, strides=2, padding='same', kernel_initializer=tf.contrib.layers.xavier_initializer_conv2d())
        bn2 = tf.layers.batch_normalization(x2, training=is_train)
        relu2 = tf.maximum(alpha * bn2, bn2)
        drop2 = tf.layers.dropout(relu2, rate=0.2)
        
        # Convolution Layer 3
        x3 = tf.layers.conv2d_transpose(drop2, 256, 5, strides=2, padding='same', kernel_initializer=tf.contrib.layers.xavier_initializer_conv2d())
        bn3 = tf.layers.batch_normalization(x3, training=is_train)
        relu3 = tf.maximum(alpha * bn3, bn3)
        drop3 = tf.layers.dropout(relu3, rate=0.2)
        
        # Output Layer 
        logits = tf.layers.conv2d_transpose(drop3, out_channel_dim, 5, strides=1, padding='same', kernel_initializer=tf.contrib.layers.xavier_initializer_conv2d())
        out = tf.tanh(logits)
        
    return out

"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_generator(generator, tf)

Tests Passed

Loss¶

Implement model_loss to build the GANs for training and calculate the loss. The function should return a tuple of (discriminator loss, generator loss). Use the following functions you implemented:

discriminator(images, reuse=False)
generator(z, out_channel_dim, is_train=True)

def model_loss(input_real, input_z, out_channel_dim):
    """
    Get the loss for the discriminator and generator
    :param input_real: Images from the real dataset
    :param input_z: Z input
    :param out_channel_dim: The number of channels in the output image
    :return: A tuple of (discriminator loss, generator loss)
    """
    g_model = generator(input_z, out_channel_dim, is_train=True)
    d_model_real, d_logits_real = discriminator(input_real)
    d_model_fake, d_logits_fake = discriminator(g_model, reuse=True)

    # One-sided Label Smoothing
    d_loss_real = tf.reduce_mean(tf.nn.sigmoid_cross_entropy_with_logits(logits=d_logits_real, labels=tf.ones_like(d_model_real) * 0.9))
    d_loss_fake = tf.reduce_mean(tf.nn.sigmoid_cross_entropy_with_logits(logits=d_logits_fake, labels=tf.zeros_like(d_model_fake)))

    d_loss = d_loss_real + d_loss_fake
    g_loss = tf.reduce_mean(tf.nn.sigmoid_cross_entropy_with_logits(logits=d_logits_fake, labels=tf.ones_like(d_model_fake)))
    
    return d_loss, g_loss

"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_model_loss(model_loss)

Tests Passed

Optimization¶

Implement model_opt to create the optimization operations for the GANs. Use tf.trainable_variables to get all the trainable variables. Filter the variables with names that are in the discriminator and generator scope names. The function should return a tuple of (discriminator training operation, generator training operation).

def model_opt(d_loss, g_loss, learning_rate, beta1):
    """
    Get optimization operations
    :param d_loss: Discriminator loss Tensor
    :param g_loss: Generator loss Tensor
    :param learning_rate: Learning Rate Placeholder
    :param beta1: The exponential decay rate for the 1st moment in the optimizer
    :return: A tuple of (discriminator training operation, generator training operation)
    """
    t_vars = tf.trainable_variables()
    d_vars = [var for var in t_vars if var.name.startswith('discriminator')]
    g_vars = [var for var in t_vars if var.name.startswith('generator')]
    
    for t in t_vars:
        assert t in d_vars or t in g_vars
    
    with tf.control_dependencies(tf.get_collection(tf.GraphKeys.UPDATE_OPS)):
        d_train_opt = tf.train.AdamOptimizer(learning_rate, beta1=beta1).minimize(d_loss, var_list=d_vars)
        g_train_opt = tf.train.AdamOptimizer(learning_rate, beta1=beta1).minimize(g_loss, var_list=g_vars)
    
    return d_train_opt, g_train_opt

"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_model_opt(model_opt, tf)

Tests Passed

Neural Network Training¶

Show Output¶

Use this function to show the current output of the generator during training. It will help you determine how well the GANs is training.

"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
import numpy as np

def show_generator_output(sess, n_images, input_z, out_channel_dim, image_mode):
    """
    Show example output for the generator
    :param sess: TensorFlow session
    :param n_images: Number of Images to display
    :param input_z: Input Z Tensor
    :param out_channel_dim: The number of channels in the output image
    :param image_mode: The mode to use for images ("RGB" or "L")
    """
    cmap = None if image_mode == 'RGB' else 'gray'
    z_dim = input_z.get_shape().as_list()[-1]
    example_z = np.random.uniform(-1, 1, size=[n_images, z_dim])

    samples = sess.run(
        generator(input_z, out_channel_dim, False),
        feed_dict={input_z: example_z})

    images_grid = helper.images_square_grid(samples, image_mode)
    pyplot.imshow(images_grid, cmap=cmap)
    pyplot.show()

Train¶

Implement train to build and train the GANs. Use the following functions you implemented:

model_inputs(image_width, image_height, image_channels, z_dim)
model_loss(input_real, input_z, out_channel_dim)
model_opt(d_loss, g_loss, learning_rate, beta1)

Use the show_generator_output to show generator output while you train. Running show_generator_output for every batch will drastically increase training time and increase the size of the notebook. It's recommended to print the generator output every 100 batches.

def train(epoch_count, batch_size, z_dim, learning_rate, beta1, get_batches, data_shape, data_image_mode):
    """
    Train the GAN
    :param epoch_count: Number of epochs
    :param batch_size: Batch Size
    :param z_dim: Z dimension
    :param learning_rate: Learning Rate
    :param beta1: The exponential decay rate for the 1st moment in the optimizer
    :param get_batches: Function to get batches
    :param data_shape: Shape of the data
    :param data_image_mode: The image mode to use for images ("RGB" or "L")
    """
    steps = 0

    input_real, input_z, learning_rate_ = model_inputs(data_shape[1], data_shape[2], data_shape[3], z_dim)
    d_loss, g_loss = model_loss(input_real, input_z, data_shape[3])
    d_opt, g_opt = model_opt(d_loss, g_loss, learning_rate, beta1)
    
    with tf.Session() as sess:
        sess.run(tf.global_variables_initializer())
        for epoch_i in range(epoch_count):
            for batch_images in get_batches(batch_size):
                steps += 1
                # Normalize
                batch_images = batch_images * 2
                batch_z = np.random.uniform(-1, 1, size=(batch_size, z_dim))
                _ = sess.run(d_opt, feed_dict={input_real: batch_images, input_z: batch_z, learning_rate_: learning_rate})
                _ = sess.run(g_opt, feed_dict={input_real: batch_images, input_z: batch_z, learning_rate_: learning_rate})
                
                if steps % 25 == 0:
                    d_train_loss = d_loss.eval({input_z: batch_z, input_real: batch_images})
                    g_train_loss = g_loss.eval({input_z: batch_z})

                    print("Epoch Number: {}/{},".format(epoch_i + 1, epochs),
                          "Discriminator Loss: {:.6f},".format(d_train_loss),
                          "Generator Loss: {:.6f}".format(g_train_loss))
                    
                if steps % 100 == 0:
                    _ = show_generator_output(sess, batch_size, input_z, data_shape[3], data_image_mode)

MNIST¶

Test your GANs architecture on MNIST. After 2 epochs, the GANs should be able to generate images that look like handwritten digits. Make sure the loss of the generator is lower than the loss of the discriminator or close to 0.

batch_size = 32
z_dim = 100
learning_rate = 0.0002
beta1 = 0.5

"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
epochs = 2

mnist_dataset = helper.Dataset('mnist', glob(os.path.join(data_dir, 'mnist/*.jpg')))
with tf.Graph().as_default():
    train(epochs, batch_size, z_dim, learning_rate, beta1, mnist_dataset.get_batches,
          mnist_dataset.shape, mnist_dataset.image_mode)

Epoch Number: 1/2, Discriminator Loss: 0.467316, Generator Loss: 15.473387
Epoch Number: 1/2, Discriminator Loss: 0.334143, Generator Loss: 8.315475
Epoch Number: 1/2, Discriminator Loss: 0.330052, Generator Loss: 6.690515
Epoch Number: 1/2, Discriminator Loss: 0.341298, Generator Loss: 5.414697

Epoch Number: 1/2, Discriminator Loss: 0.504148, Generator Loss: 8.479006
Epoch Number: 1/2, Discriminator Loss: 0.344107, Generator Loss: 5.904410
Epoch Number: 1/2, Discriminator Loss: 0.341262, Generator Loss: 6.395106
Epoch Number: 1/2, Discriminator Loss: 0.341753, Generator Loss: 5.110572

Epoch Number: 1/2, Discriminator Loss: 1.723566, Generator Loss: 1.046442
Epoch Number: 1/2, Discriminator Loss: 0.821478, Generator Loss: 1.361414
Epoch Number: 1/2, Discriminator Loss: 1.933900, Generator Loss: 0.333970
Epoch Number: 1/2, Discriminator Loss: 1.119673, Generator Loss: 0.864484

Epoch Number: 1/2, Discriminator Loss: 0.914008, Generator Loss: 1.058407
Epoch Number: 1/2, Discriminator Loss: 0.788810, Generator Loss: 1.408045
Epoch Number: 1/2, Discriminator Loss: 1.599939, Generator Loss: 0.439738
Epoch Number: 1/2, Discriminator Loss: 1.174425, Generator Loss: 1.565574

Epoch Number: 1/2, Discriminator Loss: 1.099637, Generator Loss: 1.062732
Epoch Number: 1/2, Discriminator Loss: 1.089863, Generator Loss: 1.031631
Epoch Number: 1/2, Discriminator Loss: 1.270926, Generator Loss: 0.741286
Epoch Number: 1/2, Discriminator Loss: 1.130482, Generator Loss: 0.823779

Epoch Number: 1/2, Discriminator Loss: 1.218867, Generator Loss: 1.011615
Epoch Number: 1/2, Discriminator Loss: 1.239703, Generator Loss: 0.736666
Epoch Number: 1/2, Discriminator Loss: 1.215778, Generator Loss: 0.836982
Epoch Number: 1/2, Discriminator Loss: 1.086749, Generator Loss: 1.217854

Epoch Number: 1/2, Discriminator Loss: 1.193182, Generator Loss: 0.900956
Epoch Number: 1/2, Discriminator Loss: 1.113579, Generator Loss: 0.886997
Epoch Number: 1/2, Discriminator Loss: 1.347583, Generator Loss: 0.599193
Epoch Number: 1/2, Discriminator Loss: 1.286311, Generator Loss: 0.693744

Epoch Number: 1/2, Discriminator Loss: 1.166610, Generator Loss: 0.724490
Epoch Number: 1/2, Discriminator Loss: 1.211167, Generator Loss: 0.636049
Epoch Number: 1/2, Discriminator Loss: 1.052225, Generator Loss: 0.797577
Epoch Number: 1/2, Discriminator Loss: 1.260475, Generator Loss: 0.697835

Epoch Number: 1/2, Discriminator Loss: 1.078156, Generator Loss: 0.956457
Epoch Number: 1/2, Discriminator Loss: 1.016228, Generator Loss: 1.031387
Epoch Number: 1/2, Discriminator Loss: 1.196322, Generator Loss: 0.852899
Epoch Number: 1/2, Discriminator Loss: 1.121091, Generator Loss: 1.323397

Epoch Number: 1/2, Discriminator Loss: 1.224272, Generator Loss: 1.444357
Epoch Number: 1/2, Discriminator Loss: 0.952668, Generator Loss: 1.101959
Epoch Number: 1/2, Discriminator Loss: 0.855413, Generator Loss: 1.516104
Epoch Number: 1/2, Discriminator Loss: 1.283359, Generator Loss: 0.602516

CelebA¶

Run your GANs on CelebA. It will take around 20 minutes on the average GPU to run one epoch. You can run the whole epoch or stop when it starts to generate realistic faces.

batch_size = 32
z_dim = 100
learning_rate = 0.0002
beta1 = 0.5

"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
epochs = 1

celeba_dataset = helper.Dataset('celeba', glob(os.path.join(data_dir, 'img_align_celeba/*.jpg')))
with tf.Graph().as_default():
    train(epochs, batch_size, z_dim, learning_rate, beta1, celeba_dataset.get_batches,
          celeba_dataset.shape, celeba_dataset.image_mode)

Epoch Number: 1/1, Discriminator Loss: 0.392086, Generator Loss: 4.445734
Epoch Number: 1/1, Discriminator Loss: 0.364010, Generator Loss: 3.858790
Epoch Number: 1/1, Discriminator Loss: 0.381803, Generator Loss: 4.466072
Epoch Number: 1/1, Discriminator Loss: 0.533605, Generator Loss: 2.178289

Epoch Number: 1/1, Discriminator Loss: 1.143304, Generator Loss: 0.714261

Epoch Number: 1/1, Discriminator Loss: 2.120235, Generator Loss: 0.262274
Epoch Number: 1/1, Discriminator Loss: 1.067625, Generator Loss: 1.024875
Epoch Number: 1/1, Discriminator Loss: 1.218306, Generator Loss: 1.033478
Epoch Number: 1/1, Discriminator Loss: 1.839305, Generator Loss: 2.377907

Epoch Number: 1/1, Discriminator Loss: 1.529331, Generator Loss: 0.972042
Epoch Number: 1/1, Discriminator Loss: 1.395990, Generator Loss: 1.443630
Epoch Number: 1/1, Discriminator Loss: 0.779487, Generator Loss: 1.472937
Epoch Number: 1/1, Discriminator Loss: 1.069013, Generator Loss: 0.884858

Epoch Number: 1/1, Discriminator Loss: 0.734883, Generator Loss: 1.758617
Epoch Number: 1/1, Discriminator Loss: 0.657897, Generator Loss: 1.838807
Epoch Number: 1/1, Discriminator Loss: 0.654737, Generator Loss: 2.651168
Epoch Number: 1/1, Discriminator Loss: 1.273237, Generator Loss: 1.095425

Epoch Number: 1/1, Discriminator Loss: 1.403040, Generator Loss: 0.591163
Epoch Number: 1/1, Discriminator Loss: 0.904191, Generator Loss: 1.067406
Epoch Number: 1/1, Discriminator Loss: 1.301295, Generator Loss: 0.550136
Epoch Number: 1/1, Discriminator Loss: 1.049567, Generator Loss: 1.510413

Epoch Number: 1/1, Discriminator Loss: 1.254967, Generator Loss: 0.919051
Epoch Number: 1/1, Discriminator Loss: 1.198386, Generator Loss: 0.682606
Epoch Number: 1/1, Discriminator Loss: 0.812325, Generator Loss: 1.733963
Epoch Number: 1/1, Discriminator Loss: 1.036617, Generator Loss: 0.895932

Epoch Number: 1/1, Discriminator Loss: 0.777599, Generator Loss: 1.286172
Epoch Number: 1/1, Discriminator Loss: 0.723453, Generator Loss: 1.521292
Epoch Number: 1/1, Discriminator Loss: 0.952771, Generator Loss: 1.184614
Epoch Number: 1/1, Discriminator Loss: 1.057270, Generator Loss: 1.860277

Epoch Number: 1/1, Discriminator Loss: 0.641332, Generator Loss: 3.028922
Epoch Number: 1/1, Discriminator Loss: 0.649189, Generator Loss: 2.403075
Epoch Number: 1/1, Discriminator Loss: 0.723671, Generator Loss: 1.793935
Epoch Number: 1/1, Discriminator Loss: 1.200105, Generator Loss: 1.688563

Epoch Number: 1/1, Discriminator Loss: 0.673672, Generator Loss: 2.044986
Epoch Number: 1/1, Discriminator Loss: 0.947160, Generator Loss: 1.257848
Epoch Number: 1/1, Discriminator Loss: 1.216169, Generator Loss: 0.638209
Epoch Number: 1/1, Discriminator Loss: 0.969704, Generator Loss: 0.995473

Submitting This Project¶

When submitting this project, make sure to run all the cells before saving the notebook. Save the notebook file as "dlnd_face_generation.ipynb" and save it as a HTML file under "File" -> "Download as". Include the "helper.py" and "problem_unittests.py" files in your submission.